Kubernetes scale up

[arthvpatel]

Just wondering if this is on the road map somewhere when it is stable to separate database and processing, I moved from an i9 to 3 kubernetes nodes with i5 processors for high avability on a lot of the services at home, it feels like nzbdav is struggling when it has to extract the files, in the future if it is possible to have the database separated so you can have multiple instances of nzbdav to download / provide service while keeping one database.

[nzbdav-dev]

I also use 3-node kubernetes and have definitely toyed with the Idea of HA deployments (separated frontend/backend/db, with the possibility of multiple backend replicas). Probably not anytime soon though!

it feels like nzbdav is struggling when it has to extract the files

Could you tell me a little more about this?

[arthvpatel]

On the i9-1400k processor and everything is on nzbdav nzbdav is amazingly fast, the 3 mini pcs are all identical, i5-12500t 1x512 gb nvme. Longhorn storage replicating to all the nodes and they are all rwx. If I run nzbdav on baremetal on Ubuntu server, it is decently fine but in kubernetes and longhorn, the cpu is already used for replication to other nodes, on glances the node with nzbdav shows 5.0 in 15min avg.

In my mind thinking if database is separate, I can have 2 instances of the front end which does the extraction, this would be really beneficial as it can offload extractions for each nzb on another instance

[sladg]

Hey! Would it also be able to provide a K8s yaml? I'm having a difficulty getting rclone to work properly. Would be helpful to have a working example

[nzbdav-dev]

I'm just using kube-rclone.

I've also tried csi-rclone ( a few years ago), and it worked well too, but I found kube-rclone to be a bit simpler.